BestBdStop for feasibility problems
AnsweredHello,
I am fine-tuning my model for a feasibility problem.
I use two different ways : create a feasibility problem (minimize 0) or minimize an objective but I set the solution limit attribute to 1 to allow some heuristics and maybe find a feasible solution faster.
For both cases, I asked myself : is it possible to improve the computation time by setting a best bound stop ? When I try, it always returns the Objective Limit status, but is it guaranteed that a feasible solution had been found ? And if it's the case, how to print the variables that leads to a feasible solution ?
In addition, is there any hint to improve the computation time for a non-covex MIP feasibility problem ? I try with MIPFocus, FeasibilityTol and ImproveStartTime, but are there any other parameter eligible to improve the model ?
Thanks a lot,
Timothy
-
-
Gurobi Staff
Hi Timothy,
You can query the model attribute SolCount at the end of the optimization run to check if a feasible solution was found. If model.SolCount >0, you can query the variable attribute X to find its value in the best solution found.
For performance, you might want to experiment with the NoRelHeurTime and Heuristics parameters.
Best regards,
Simran0 -
-
Thank's a lot !
Now I have one more question. Does this also works if the model in supposed to be infeasible ? Which parameters could improve the solving time of an infeasible model ?
0 -
-
-
Hi Timothy,
You can check the status attribute of the model at the end of the optimize call to check if the model is infeasible.
Which parameters could improve the solving time of an infeasible model ?
Do you mean the running time of the computeIIS method that finds the smallest subset of constraints and variables that make the model infeasible? For that, please take a look at the parameters and attributes: IISMethod, IISConstrForce, IISLBForce, and IISUBForce.
- Simran
0 -
Please sign in to leave a comment.
Comments
3 comments